Redesigning ONYX around daily fund operations

ONYX is the ETF order-management platform used by the 21Shares team. I led the V2 redesign in 2024 with the CTO and three engineers, from research and workflow mapping through the complete light and dark screen set. The product brought NAV, PCF, fund accounting, orders, rebalancing and exceptions into one daily operating environment. The platform processed more than $8.5B in creation and redemption orders across 50+ products, as reported by Filip Geschwandtner, who managed product design at the time.

The existing product made routine work harder to follow. Components were duplicated, forms behaved inconsistently and important steps were easy to miss. Unclear status, ownership and exceptions made multi-step work harder to manage. I focused the redesign on one interface system and a dashboard that put urgent work, lifecycle status and ownership before routine tasks.

Planning delivery with the engineering team

The redesign had to progress while engineers continued developing ONYX. I planned the roadmap and milestones with the CTO and engineering team, using the workflow map and component audit to decide what needed to be resolved first. The sequence covered the shared component foundation, the core journeys and the screen sets engineering needed for implementation.

I ran sprint planning, stand-ups, retrospectives and stakeholder reviews, and established weekly design and delivery rituals. These meetings connected the next milestone to the work in progress, giving us regular opportunities to review priorities, surface blockers and work through design or implementation questions. I guided the team towards the agreed milestones and deadlines while staying involved in the detailed interface work and the constraints engineers were facing.

Understanding the work before changing the product

I interviewed power users across NAV, PCF, rebalancing, fund accounting, orders and exceptions. They showed me where unclear status, ownership and handoffs slowed the work down, which gave us an order for the navigation and dashboard changes. I also mapped the recurring fund cycles behind the interface, exposing dependencies the old navigation had hidden.

Engineering was still working on the product, so I kept the redesign staged. I mapped the legacy workflows and components first, then stabilised the component library before redesigning and testing the core journeys in working prototypes. The first changes stayed close to existing behaviour, while the operating model showed us where the product structure needed to change.

Building one system for both themes

The audit found duplicate components, unused styles and inconsistent states. I consolidated variants and sizes around the patterns that repeated across workflows, then aligned colour, spacing, typography, elevation and state tokens across the light and dark themes. I mapped the shared framework defaults into the design system so the engineers needed fewer one-off solutions.

I documented component anatomy, variants, sizes, usage rules and the hover, focus, selected, processing, success and error states. Each handoff combined behaviour, measurements and implementation notes. This gave the engineers a clear specification for tables, forms, navigation and notifications while keeping the work compatible with the frameworks already used in the product.

Applying the system to the daily workflow

I redesigned orders, product management, allocation and rebalancing around the information operators needed at each step. Product information, settings and rebalancing actions stayed together. Order flows used the same lifecycle and status patterns, and I organised dense allocation data around the decisions operators needed to make. The dashboard placed work that could block a fund or deadline at the top, with its owner and current status beside it.

I delivered the documented dual-theme framework, the full V2 screen set and a dashboard organised around daily priorities. The work covered the operating model, components and workflow screens while the engineers continued building the product.

Faster ETF creation and fewer support requests

Following the redesign, ETF creation was approximately 67% faster, support tickets fell by around 39%, and PCF errors could be detected in minutes rather than hours. These team results were reported by Filip Geschwandtner, who managed product design at the time.